[Registration of intraoperative 3D ultrasound with preoperative MRI data for computer-assisted orthopaedic surgery]

The trend towards minimally invasive surgery leads to an increasing demand for methods enhancing in situ spatial orientation. Hence, navigation systems are being more and more used for many operations. At the beginning of a navigated surgery the coordinate systems of the patient need to be registered with the navigation system. Our registration method is based on intraoperative freehand ultrasound imaging and overcomes some problems related to landmark-based registration, such as invasiveness, inaccuracy and the time-consumption of the process. After a successful registration, the operation can be conducted safely with the aid of the navigation system which provides enhanced spatial orientation on high-quality preoperative image data (e.g., MRI, CT). Soft tissue is depicted especially well in MRI data. Since imaging of soft tissue close to the joints is of high diagnostic importance, MRI scans are often part of the preoperative routine. These data can also be utilised by a navigation system. In this article we present our first results on the registration of 3D ultrasound and MRI data of the knee.

3D ultrasound and T (1)/T (2)-weighted MRI data were acquired from the distal femur and the proximal tibia. The ultrasound volume data were pre-processed to enhance the contrast of bone and surrounding tissue. A dataset combined from T (1)- and T (2)-weighted MRI images was also created. Points of the bone surface corresponding optimally to the bone surface demonstrated in the ultrasound images were extracted from the MRI volume data. For the surface extraction of the femur the T (1)-weighted image data were used, and for the tibia the combined MRI dataset was used. Based on the extracted bone surface, MRI and ultrasound volume data were registered with a surface-volume registration algorithm.

Registrations of ultrasound data of both the femur and the tibia were successful. Bone and cartilage was depicted well in the ultrasound data and visual inspection showed the accurate overlay of bone structures in MRI and ultrasound volumes.

We have demonstrated the feasibility of registering bone structures based on 3D-ultrasound and MRI data. Our method allows for minimally invasive routine surgery of orthopaedic diseases which so far had to be operated conventionally because other methods for navigated surgery are too expensive, inaccurate or time-consuming.